On a Tuesday in mid-March, as yet another ChatGPT fabrication made the rounds on LinkedIn, a London startup announced it had raised a quarter-million pounds to do something both obvious and surprisingly rare: check whether AI-generated content is actually true.
YAi Digital's £250,000 pre-seed round, led by Jenson Ventures and completed under the UK's SEIS scheme, arrives at a curious inflection point. Everyone knows AI hallucinates. Study after study confirms it—rates consistently land somewhere between 40% and 60%, depending on who's measuring and how. Yet the content keeps flowing, and someone still has to verify it, paragraph by paragraph, claim by claim.
That verification gap is what YAi wants to occupy.
The Problem No One's Quite Solved
Consider the numbers from NewsGuard's latest AI Misinformation Monitor, released this March. Leading chatbots—the ones millions of people consult daily—delivered accurate information just 58.5% of the time when tested against a controlled set of false narratives. That's barely better than a coin flip. In December, the failure rate spiked to 62%.
These aren't fringe edge cases. We're talking about systems that power marketing copy, customer service scripts, internal reports—the everyday infrastructure of modern business communication. And someone, somewhere, is supposed to catch the errors before they compound.
Enter ContentLift, YAi's flagship product. It doesn't generate content. It evaluates what other tools produce: fact-checking sources, flagging unsupported assertions, scoring tone and readability. Think of it as a quality-control checkpoint between your AI draft and your audience—a role that, until recently, required a human with a browser tab open and a growing sense of paranoia.
The pitch is straightforward, maybe deceptively so. For agencies churning out AI-assisted blog posts and brand content at scale, ContentLift promises to automate the verification step that's become the real bottleneck. The company launched the product on Product Hunt last September, collecting 196 upvotes and a perfect 5.0 rating from nine early reviewers. A seven-day free trial keeps the barrier to entry low.
A Team That's Been Here Before
YAi's founding trio isn't new to the pressures of scaling a business. CEO Yana Lapitskaya and COO/CMO Jessy Conflon previously built and exited Yay!Starter Marketing, a UK agency that carved out a niche in e-learning, fintech, and ecommerce before winning Small Business of 2021 honors. CTO Vyacheslav Lukin brings technical credibility from a prior exit—a computer-vision AI platform for transport management.
They incorporated YAi in December 2023, which means the company is barely 15 months old. Already, they've landed on BusinessCloud's MediaTech 50 list (announced in May 2025) and run an October hackathon with partners like SkilledUp Life to test the product in the wild. Speed, it seems, is part of the strategy.
But speed only gets you so far when the competition includes better-capitalized rivals. Originality.ai markets an automated fact-checker with real-time verification. Writer.com, backed by enterprise clients, touts built-in claim validation powered by knowledge graphs. Academic labs continue publishing new detection methods—FactSelfCheck, HaluCheck, FacTool—though few have made the leap from research paper to revenue.
The £250,000 Question
Whether a quarter-million pounds provides enough runway to refine the technology, build meaningful distribution, and carve out defensible market share is the real test. YAi operates from a shared workspace on Paul Street in East London with a team sized between one and ten employees—lean even by startup standards.
The money will need to stretch across product development, customer acquisition, and the cost of competing against incumbents with deeper pockets and established brand trust. It's a classic pre-seed puzzle: prove the concept works, fast, before the capital runs dry.
YAi's wager is that the bottleneck in AI content isn't generation—it's trust. Anyone can produce a thousand blog posts a week now. The hard part, the part that still requires humans staring at screens and second-guessing every statistic, is making sure those posts won't embarrass you later.
If Lapitskaya and her team are right, there's a sizable market for a product that simply makes sure the robots aren't lying. If they're wrong—or if the majors solve hallucination faster than expected—then ContentLift becomes one more quality-control tool in a crowded field, racing against time and a modest bank balance.
For now, at least, someone's willing to bet £250,000 that verification is the problem worth solving. Whether that proves prescient or premature will depend on how quickly trust in AI-generated content continues to erode—and whether anyone can build it back.
